Dexfenfluramine, Fluoxetine, and Weight Loss ... - Semantic Scholar

201

NEUROPSYCHOPHARMACOLOGY 1993-VOL.9, NO.3

Dexfenfluramine, Fluoxetine, and Weight Loss Among Female Carbohydrate Cravers Judith Wurtman, Ph.D., Richard Wurtman, M.D., Eliot Berry, M.D., Ray Gleason, Ph.D., Hannah Goldberg, B.A., Janine McDermott, B.S., Merton Kahne, M.D., and Rita Tsay, M.S.

The consumption of excess calories as carbohydrates (CHO)-rich, protein-poor snacks characterizes the overeating of obese CHO cravers, premenstrual women, patients with Seasonal Affective Disorder, and former smokers. This specific appetite for CHOs may involve m-ain serotonin, as the synthesis and release of this neurotransmitter can increase following consumption of CHG-rich foods. To examine whether weight loss produced by serotoninergic drugs involves a selective reduction in CHO intake, obese females who consumed at least 30% of their daily calories from CRO-rich snacks were treated with dexfenfluramine ({DFJ 15 mg b.i.d.); ftuoxetine ({FLJ 20 mg t.i.d.); or placebo (PL) for 12 weeks. Weekly weight loss for 25 of 29 PL com pieters was

KEY WORDS: Fluoxetine; Dexfenfluramine; Obesity; Appetite; Carbohydrate

Dexfenfluramine (OF) is utilized in many countries for the treatment of human obesity, particularly that as­ sociated with the consumption of excessive amounts

of carbohydrate (CHO)-rich foods (Ditschuneit et al.

1991; Guy-Grand et al. 1989; Wurtman et al. 1985). Fluoxetine (FL), currently in use as an antidepressant drug, has recently been found to be an effective anorec­

tic agent as well (Ferguson and Feighner 1987; Kut­ nowski et al.

1987).

1990; Levine et al. 1987; Wise 1992; Zerbe

Both drugs act to enhance serotonin-mediated

0.22 kg ± 0.06 (mean ± SEM); for 21 of 28 DF completers, 0.56 ± 0.08 kg; and for 18 of 30 FL completers, 0.58 ± 0.09 kg (PL < DF FL; p .039). Seven FL subjects, 2 PL subjects, and 1 DF subject =

=

withdrew from the study due to side effects; other withdrawals were due to intercurrent illness or personal problems. Prior to treatment, subjects consumed over 40% of their daily CRO intake from snacks. Both of the drugs selectively decreased CRO snack intake (p < 0.05); DF, but not FL, also decreased meal CRO intake (p < .025). These results suggest that weight loss following treatment with serotoninergic drugs may relate to a selective decrease in CRO appetite. [Neuropsychopharmacology 9:201-210, 1993]

brain neurotransrnission by blocking brain serotonin reuptake; OF through its metabolite, nordexfenflura­ mine, also releases serotonin into synapses. It has been suggested that the weight loss produced by these drugs may be related to their effect in spe­ ciftcally decreasing the excessive intake of CHO-rich foods (Wurtman et al. 1985; Ferguson and Feighner 1987). Because such foods are usually rich in fat, reduc­ ing CHO intake can account for a substantial reduction

1985; Ferguson 1987). Brain serotonin appears to influ­

in total calorie intake (Wurtman et al. and Feighner

ence CHO consumption; evidence from studies in which animals or human volunteers were treated with dietary or pharmacologic interventions that increase

From the Clinical Research Center, Massachusetts Institute of Technology, Cambridge, Massachusetts. Address correspondence to: Judith J. Wurtman, Ph.D., Clinical Research Center E17-438, Massachusetts Institute of Technology, Cambridge, MA 02139. Received November 24, 1992; revised March 15, 1993 and May 26, 1993; accepted May 31, 1993.

serotonin-mediated neurotransrnission indicates selec­ tive decreases in consumption of CHO-rich foods (Blun­ dell 1992; Leibowitz and Shor-Posner 1987; Luo and Li 1991; Wurtman et al. 1983; Wurtman and Wurtman 1979). When animals are treated with anorectic doses of OF or FL, the subsequent reduction in calorie intake

C 1993 American College of Neuropsychopharmacology Published by Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010

0893-133X/93/$6.00

202

NEUROPSYCHOPHARMACOLOGY 1993-VOL. 9,

J. Wurtman et al.

NO.3

is also associated with a selective decrease in CHO con­

be in good general health, a nonsmoker, and at least

sumption; protein intake remains unchanged (Luo and

3 months beyond the completion of any weight loss pro­

1991; Wurtman and Wurtman 1979). In contrast,

gram. Applicants who reported moderate-to-severe

anorectic drugs that affect catecholamine-mediated neu­

premenstrual mood and appetite changes and! or a clear

rotransmission nonselectively decrease food intake

pattern of seasonal weight gain and loss were not con­

(Moses and Wurtman 1984). Obese humans who typi­

sidered for the study because such fluctuations in mood

cally consume excessive amounts of CHO foods, par­

and appetite could potentially influence the treatment

Li

ticularly as snacks, reduce their CHO snack intake by

responses during the study. Eligible subjects were then

approximately 40% in response to OF, but they do not

evaluated at the Massachusetts Institute of Technology

significantly alter their protein consumption. Anorec­

(MIT) Clinical Research Center (CRC). The subjects

tic doses of FL might be expected to similarly affect CHO

were asked to sign a consent form that had been ap­

intake because of this drug's similarity to OF in enhanc­

proved by the MIT Committee on Use of Humans as

ing serotonin-mediated neurotransmission. However,

Experimental Subjects (COUHES) and the CRC Advi­

recent studies indicate that anorectic doses of FL may

sory Committee. They then underwent examination by

affect catecholaminergic as well as serotoninergic neu­

a physician that included a history and physical exami­

1992; Grignaschi and

nation, urinalysis, electrocardiogram, thyroid profIle,

Samanin in press) by inhibiting catecholamine reup­

HGC, complete blood count, and blood profIle (BPRO:

rotransmission (Garattini et al.

take; moreover, rats treated with metergoline, a sero­

glucose, blood urea nitrogen, creatinine, total protein,

tonin-receptor antagonist, still respond to the anorec­

albumin, calcium, phosphorous, total bilirubin, uric

1992; 1989). If FL-mediated anorexia in humans

acid, cholesterol, triglyceride, sodium, potassium, chlo­

also involves catecholaminergic mechanisms, then the

hydrogenase, serum glutamic, oxaloacetic transami­

drug'S influence on macronutrient intake may differ

nase, serum glutamic-pyruvic transaminase).

tic actions of FL but not OF (Garattini et al. Samanin et al.

ride, total carbon dioxide, alkaline phosphate, lactic de­

from that of OF. Little is known from clinical trials about specific changes in food intake following FL adminis­ tration. One study did investigate changes in appetite;

Baseline Food Intake

the results indicated a lessening of appetite for CHO­

One hundred and two subjects who were in good phys­

rich foods among individuals treated with FL but not

ical health (as determined by physical examination,

with placebo (Ferguson and Feighner 1987). However,

blood chemistry, and electrocardiogram) and whose

that study was not designed to measure food intake

weight status met protocol criteria were asked to par­

directly and relied instead on subjective perception. The

ticipate in a 4-day outpatient measurement of their meal

present study compares the effects of OF and FL treat­

and snack intakes to identify those whose food intake

ments on calorie, macronutrient intake, and weight loss

reflected a consistent appetite for CHO-rich snack

among obese women. The subject population was re­

foods. Previous studies have shown that most obese

stricted to individuals who claimed to consume exces­

subjects who claimed that their consumption of sweet

sive amounts of CHO-rich foods as snacks ("CHO

and starchy snacks was associated with an inability to

cravers"), and whose pretreatment pattern of food in­

lose weight did in fact eat at least 30% of their total daily

take confirmed this claim.

calories in the form of between-meal CHO-rich snacks (Wurtman et al.

1987).

A two-phase approach was used to identify sub­ METHODS AND SUBJECTS Subject Recruitment and Screening

jects who would meet our criteria for CHO craving. Prior to admission to the screening component of the study, interested subjects fIlled out a mailed question­

Subjects were recruited through newspaper, radio, and

naire detailing their typical pattern of snack intake. This

mass transit advertisements directed at women who

questionnaire asked them if they snacked daily and,

40 lbs to 60 lbs above ideal body weight, accord­

if so, to indicate when during the day they snacked.

were

ing to the Metropolitan Life Insurance Height and

The subjects indicated, from a long list of protein-,

Weight Table for midpoint frames (Metropolitan Height

CHO-, and fat-rich snacks as well as fruits and vegeta­

1983), and who perceived their

bles, which snacks they were most likely to eat. The

obesity as being sustained by an excessive appetite for

questionnaire defmed "snacking" as eating between or

and Weight Tables

sweet and starchy snack foods. Potential subjects were

after meals when subjects were not hungry. Subjects

asked to complete and return questionnaires on their

who indicated that they snacked daily on only CHO­

health history, menstrual symptomatology, weight-loss

rich foods were then asked to participate in the base­

history, annual patterns of weight gain and loss, and

line measurement procedure.

typical daily snack patterns. Acceptance into the out­

The dietary department of the CRC provided the

patient screening procedure required the applicant to

subjects with an assortment of isocaloric, low-fat,

Dexfenfluramine, Fluoxetine, and Weight Loss 203

NEUROPSYCHOPHARMACOLOGY 1993-VOL. 9, NO.3

Table 1.

Meal and Snack Foods Protein

CHa

Fat

( g)

KCal

(g)

(g)

(g)

% KCaliFat

43 37 143 105 105 57 128 106 135 111

150 147 150 150 142 150 140 150 154 150

4 3 17.5 4 22 4 23 26 5 23

28 18 3 25 0 36 2 2 28 0

4 7 6.25 4 5 2 3 4 3 6.75

24 43 38 24 32 12 19 24 18 40

27

100 120 110 100 80 106 140

1 1 3 13 7 1 3

24

0 3 2 5 5 2 6

Amount

Meal item Oatbran fruit bar Cinnamon raisin roll Cottage cheese with chives Pasta salad Seafood salad Soft roll Tuna with diet dressing Chicken with BBQ sauce Stuffed potato with cheese Cod with cheese Snack item Cereal (Frootloops) Cereal (QuakerOH's) Pretzel sticks Luncheon meat Cheese Fig bars Goldfish crackers (cheese)

44

28 71 28 30 30

protein-rich, and CHO-rich foods in excess of what they might consume (Table 1). Meal foods contained an av­ erage of 32% Kcal from fat, and snack foods contained 26% Kcal from fat. The foods were packaged in preweighed, easily transported containers for con­ sumption away from home. The meal and snack food items provided by the study were relatively low in fat to make them compatible with the objectives of a weight-loss study. Subjects were told to eat spontane­ ously, not to restrict their food intake, and to consume ony l and the foods designated as snack foods for in-between meal snacks. A simple log book was provided in which the subject could check off the foods actually consumed against a printed list and could note the time of day when each food was eaten. The log book was designed such that the subjects could record whether they ate a particular food item as a meal or snack food, as well as the time of food consumption. Because it was possi­ ble that they would also consume foods other than those provided by the study, they were asked to note their consumption of those foods as well. All food containers and all uneaten food were returned to the dietician for weighing. The subjects were weighed prior to the food intake measurement and on the day of completion. Activity Levels

Subjects kept a record of their physical activity for 4 days coincident with the period of baseline food intake mea­ surement. None of the subjects engaged in any formal exercise regime; physical activity was limited to that necessary for maintaining daily routines (i.e., walking

22

21 3 1 20 18

0

22

16 45 56 17 38

from the car to work, climbing stairs in the home, meal preparation, carrying groceries, doing laundry, etc.) Most of the subjects' waking hours were spent seated. Study Format

Subjects whose CHO intake as snacks met the protocol criteria were randomly assigned to a FL, OF, or placebo (PL) treatment group. The study was conducted in double-blind fashion; subjects received identical packets of three pills that provided a total of 60 mg FL, 30 mg OF, or PL. They were told to take one pill in the morning, one at noon, and one in the early evening; the noon pill in the OF packet was a placebo. The doses of FL and OF were those known to produce anorectic effects (Guy-Grand et al. 1989; Levine et al. 1987). Sub­ jects were given a food plan that instructed them as to the number of servings and portion sizes they should consume to obtain their total daily nutrient require­ ments. The dietary information emphasized the impor­ tance of eating a variety of foods, and subjects were given exchange lists containing examples of nutrient­ dense, low-fat foods. No additional information on weight loss was offered to the subjects. There was no discussion be­ tween the staff and subjects on exercise, behavioral modification, stressful events that might have affected weight loss, or nutritional information other than repeated admonitions to consume as little fat as possi­ ble. Because the subjects were entered individually into the study, no interaction between subjects was possi­ ble. The staff was uniformly considerate and kind to the subjects, many of whom reported after completion

204

J. Wurtrnan et al.

that weekly contact with the staff was an important posi­ tive component in their compliance with the study. Measurements

Weight, vital signs, side effects, and assessments of mood and premenstrual symptomatology were ob­ tained on the day prior to the initiation of drug treat­ ment. The presence of depressive symptoms over the previous week was assessed using the interviewer­ administered Hamilton Depression Rating Scale (HDRS) (Hamilton 1967), the self-administered ProfIle of Mood States (POMS) (McNair et al. 1971), and the Center for Epidemiological Studies-Depression (CES­ D) (Radloff 1977) test. Subjects also ftlled out the Stand­ ford Sleepiness Scale (SSS) (Hoddes et al. 1973) and the Menstrual Symptomatology Questionnaire (MSQ) (Abraham 1981). Baseline side effects were monitored by recording spontaneous descriptions of mood and physical status during the previous week. Following this, subjects were asked to report any illness or major life event that occurred during the previous week and to note the week of their menstrual cycle or, for meno­ pausal women, the week of hormone replacement ther­ apy. Blood was drawn for baseline drug levels and BPRO. Subjects were treated for 12 weeks; they returned to the CRC weekly to receive their medications. At that time, their vital signs, side effects, and weight were evaluated and measured, and they ftlled out the POMS, CES-D, SSS, and MSQ forms. The research nurse reviewed the side-effect report with each subject prior to dispensing medication for the following week. The side-effect reports were also subsequently reviewed by the study physicians. Every fourth week, blood was drawn for measurement of drug levels and BPRO, and food intake was monitored over 4 days, using the methods described above for baseline food intake. Two weeks following the end of the treatment period, subjects returned to the CRC for a posttreat­ ment assessment of drug levels and BPRO. At that time, they met with the research staff and were given infor­ mation on dietary intake and exercise. When subjects dropped out of the study, one of the research physi­ cians sent a full report of their medical and treatment status, along with their reasons for withdrawal, to the MIT COUHES. Data Analysis

Weight, food intake, and test scores were analyzed using a two-factor repeated-measures analysis of vari­ ance with time as the repeated measure. Food intake was averaged for each 4-day measurement period. Post­ hoc pairwise comparisons among study group and time means were tested for signifIcance using the Newman-

NEUROPSYCHOPHARMACOLOGY 1993-VOL. 9,

NO.3

Kews test. Relationships between variables were tested for signifIcance using Pearson's product moment corre­ lation. Differences in distributions of dropouts and side effects among study groups were compared using the chi-square statistic. Data are expressed as means with standard errors of the mean used as the measure of dispersion. Statistical signifIcance was determined at p � .05. RESULTS

Eighty-seven women qualifIed for entry into the study on the basis of data from the baseline food intake screen· ing. All of them met our criteria for CHO cravers, can· suming more than 40% of their total CHO intake as high-CHO snack foods and more than 30% of their to­ tal daily calories as such foods. (Protein intake from snacks accounted for slightly more than 20% of their total daily protein intake.) Fifteen subjects were elimi· nated during the screening period because they re­ stricted their calorie consumption during the food in­ take measurement period to under 1000 Kcal a day and admitted to the diet staff that they wanted to use the baseline period to initiate their weight loss regimen. Fat was a fIxed constituent of the meal and snack fo ods; subjects could not alter the fat contents of the foods by, for example, choosing to eat or avoid high-fat ingre­ dients such as butter or mayonnaise. Although the foods varied somewhat in the number of grams of fat they contained, the amount of fat the subject consumed was determined mainly by the number of servings of protein or CHO-rich foods that she chose to eat. How­ ever, the fat content of the foods was kept as low as possible because the subjects wowd be eating the same foods during the weight loss component of the study, and we did not want the foods used in the study to be incompatible with weight loss. The three groups of subjects were similar in age, body mass index (BMI) (Table 2) and in reporting a sedentary level of physical activity (Table 2); none showed evidence of depression on self-reported or interviewer-rated tests of mood at the time they entered the study (Table 2). Subjects were entered singly into each treatment group; 64 subjects completed the study: the PL group consisted of 25 of 29 initial subjects, OF 21 of 28 sub­ jects, and FL 18 of 30 subjects (Table 2). Weight loss per week was calculated for all subjects (Le., for those who completed the study and for those who dropped out). No difference was found in the rate of weekly weight loss between completers and dropouts at the time the latter withdrew from the study; hence, the results given represent those of the total group. The weekly rates of weight loss were similar for the two drug treatment groups and signifIcantly greater than that of the PL treatment group (p .039). The PL-treated sub=

Dexfenfluramine, Fluoxetine, and Weight Loss 205

NEUROPSYCHOPHARMACOLOGY 1993-VOL. 9, NO. 3

Table 2.

Subject Profile Treatment

Variable

Placebo

Number Age Height (cm) Weight (kg) BMI Completers Dropouts Physical Activity HDRS CES-D

Dexfenfluramine

Fluoxetine

28

30 41.0 ± 1.9 165. 0 ± 1.2 90. 1 ± 1. 9 33.1 ± 0.6 18 12 Inactive 7.4 ± 1. 4 8.89 ± 2.34

29 39.5 164.2 88.33 32.8

± ± ± ±

1. 7 0. 9 1.5 0. 5

25 7 Inactive 8. 5 ± 1.4 9.92 ± 1.23

jects lost 0.22 ± 0.06 kg per week, DF subjects lost 0 .56 ± 0.08 kg per week, and FL subjects lost 0.58 ± 0.09 kg/per week (groups include both completers and noncompleters). The cumulative weight loss (of com­ pieters) at the end of the treatment period was also simi­ lar between drug-treated subjects and signifIcantly more than among those receiving PL (p < .05) (Fig. 1). The PL-treated subjects lost 3.0 ± .74 kg, DF subjects lost 6.6 ± 0.96 kg, and FL subjects lost 7.1 ± 1.09 kg. The PL-treated subjects showed a small weight gain of ap­ proximately 0.14 ± 0.19 kg at the end of the 5th and 9th treatment weeks, and net losses at the end of all the other weeks. Because the amount of weight to be lost was similar but not identical among subjects, weight loss was also calculated as a percent of the amount of weight that needed to be lost for each subject to attain ideal body weight. The PL-treated subjects lost 11.3% of this weight, DF subjects lost 27%, and FL subjects lost 26% (Fig. 2). Food Intake

All subjects consumed more than 40% of their total CHO intake from snacks during the baseline food in___

41.2 163. 7 85. 7 32.0

± ± ± ±

1.7 1. 1 1. 7 0.5

21 4 Inactive 10.1 ± 1. 8 12.95 ± 2.77

take period. Protein intake from snacks accounted for slightly more than 20% of total daily protein intake. Daily calorie consumption was similar between all groups prior to the initiation of treatment (Table 3). Sub­ jects consumed approximately 1250 calories from meals and 790 calories from snacks during the baseline period of food intake measurement. Their calorie intake dur­ ing the study may have underrepresented what they normally consumed inasmuch as the foods supplied by the CRC were probably lower in fat than those the sub­ jects would choose to consume at home or in a restau­ rant (Table 3). All treatments resulted in a decrease in calorie in­ take from meals and snacks; the reductions, however, were significantly greater for the drug-treated groups than for PL group. Decreased CHO intake from meals and snacks accounted for the greatest effect of the drugs on food intake (Table 3). By the end of the third treat­ ment period, snack CHO intake by the PL-treated sub­ jects was decreased by 48% of baseline. The DF-treated subjects had decreased snack CHO intake by 75%, and FL-treated subjects decreased snack CHO intake by 60%. The decreases with DF and FL treatment were significantly greater than with PL (p < .05). 10

PLACEBO

OEXFENFLURAMINE - FLUOXETINE

•••••

10

WEIGHT tOO; (KG)

*

*

8

8

WEIGHT LOSS (%)

6

6 4

2

4 2 0 o

Figure 1.

c:::::::J OEXFENFLURAMINE _ FLlJOXETlNE c:::::::J Pl.ACEBO

1

2

3

4

5

6

7

8

TREATMENT WEEK

9

10

11

12

Cumulative weight loss (kg) in subjects receiving placebo (PL), dexfenfluramine ([DF] 15 mg, twice daily), and fluoxetine ([FL] 20 mg, three times daily) for 12 weeks. • p < .05 differs from PL-treated group.

Fig ure 2.

4

8

12

WEEKS ON TREATMENT

Weight loss as the percent of the amount to be lost to attain ideal body weight according to the Metropolitan Life Insurance Height/Weight Tables for Women. * p < . 05 differs from PL-treated group.

206

NEUROPSYCHOPHARMACOLOGY 1993-VOL. 9,

J. Wurtman et al.

Food Intake During Baseline Period and Following 12 Weeks of Treatment

Table 3.

Placebo

Calories Meal Baseline 1208 ± 41 Week 12 1126 ± 51 Snack Baseline 720 ± 49 Week 12 368 ± 66 Protein (g) Meal Baseline 86 ± 3 78 ± 5 Week 12 Snack 21 ± 3 Baseline Week 12 10 ± 2 CHO (g) Meal Baseline 151 ± 7 Week 12 141 ± 9 Snack Baseline 109 ± 8 Week 12 57 ± 11

Side Effects and Dropouts

Fluoxetine

Dexenfluramine

NO.3

The number of subjects who withdrew from the study due to side effects differed signifIcantly

(p

.02) be­